Content Copyright © 2023 Bloor. All Rights Reserved.
Also posted on: Bloor blogs
At time of publication, Teradata has just announced the release of Teradata AI Unlimited. At its core, this is an attempt to give data professionals and developers of all types access to the tools they need to freely experiment with AI as quickly as possible while minimising the relevant costs. More specifically, it provides them with personal, temporary, non-production environments in which to do so. The expectation is that these efforts will, over time, lead to the implementation of more and better AI within your production systems and processes.
In effect, AI Unlimited gives your data scientists, data engineers, and so on access to a personalised instance of a Teradata environment, which they can play with to their heart’s content. If/when their efforts bear fruit, they can be implemented as part of your production platform. AI Unlimited instances can be created on-demand and without a long start-uptime, essentially acting as an accelerant for your AI development efforts. What’s more, due to the temporary nature of these instances, there is no need for data persistence, which can be (and frequently is) rather costly. Similarly, because they’re non-production, you will often be able to relax what would otherwise be restrictive (and potentially obstructing) IT controls and data access limitations when working with AI Unlimited instances.
More technically, AI Unlimited is a serverless AI and ML engine that can be deployed on-demand to the cloud of your choice where your data resides. The engine itself is used for compute, and is supported by object stores that contain your data. It is most readily interacted via Jupyter, which can be used as a control plane, although there is also a command line interface available that can be connected to a tool of your choice. GitHub is used to persist metadata and schema backups; most notably, this allows the product to provide a consistent (and customisable) look and feel for each user despite each personalised instance only existing temporarily.
In addition, Teradata has collaborated with Microsoft to deliver its new offering on both Microsoft Fabric and the Azure Marketplace. In the former case, integration with the Microsoft ecosystem will be as tight as you would expect, such that you can quickly and easily leverage AI Unlimited functionality, like spinning up a new cluster, from within the Fabric platform. The Azure marketplace offering, in contrast, is self-service, and relies on more direct interaction with the infrastructure described above. In either case, AI Unlimited seem like an obvious win if you’re serious about innovating with AI.